public inbox for cygwin@cygwin.com
 help / color / mirror / Atom feed
From: Thomas Wolff <towo@towo.net>
To: cygwin@cygwin.com
Subject: Re: stty -cooked not usable since cygwin-3.1.1-1
Date: Wed, 29 Jul 2020 01:30:44 +0200	[thread overview]
Message-ID: <f935b0f6-1de3-e2d1-285e-f26cd945985a@towo.net> (raw)
In-Reply-To: <22f2becd-1f2e-5ff8-4999-b34f2fb0bff3@bahnhof.se>



Am 28.07.2020 um 21:51 schrieb Rabbe Fogelholm:
> On 2020-07-28 18:56, Thomas Wolff wrote:
>>
>>
>> Am 28.07.2020 um 18:38 schrieb Takashi Yano via Cygwin:
>>> On Tue, 28 Jul 2020 15:56:54 +0200
>>> Rabbe Fogelholm wrote:
>>>> Rabbe Fogelholm wrote:
>>>>> Takashi Yano wrote:
>>>>>> On Mon, 13 Jan 2020 11:52:43 +0100
>>>>>> Rabbe Fogelholm wrote:
>>>>>>> I am running a console Java program that is started from a 
>>>>>>> shellscript
>>>>>>> wrapper. Before invoking Java the wrapper calls `stty -cooked'. 
>>>>>>> The Java
>>>>>>> program polls the keyboard using System.in.available() and reads
>>>>>>> characters immediately using System.in.read(), without waiting 
>>>>>>> for the
>>>>>>> Enter key to be pressed.
>>>>>>>
>>>>>>> This way of combining `stty -cooked' and Java has stopped 
>>>>>>> working since
>>>>>>> version 3.1.1-1 of the Cygwin package. The Java thread that 
>>>>>>> reads the
>>>>>>> keyboard hangs until Enter is pressed, which is not desirable.
>>>>>>>
>>>>>>> I had to downgrade to version 3.0.7-1 to resolve the problem.
>>>>>>>
>>>>>>> Versioning information:
>>>>>>>
>>>>>>> java version "1.8.0_202"
>>>>>>> Java(TM) SE Runtime Environment (build 1.8.0_202-b08)
>>>>>>> Java HotSpot(TM) 64-Bit Server VM (build 25.202-b08, mixed mode)
>>>>>>>
>>>>>>> OS Name: Microsoft Windows 10 Pro
>>>>>>> Version: 10.0.17763 Build 17763
>>>>>>> System Type: x64-based PC
>>>>>>>
>>>>>>> See also the enclosed cygcheck.out.
>>>>>>>
>>>>>>> To demonstrate the issue I enclose a small Java program that 
>>>>>>> should be
>>>>>>> able to read single keystrokes when `stty -cooked' is in effect.
>>>>>> Does your java program work in command prompt? cygwin 3.1.x uses
>>>>>> pseudo console, so the native (non cygwin) program works as if it
>>>>>> is executed in command prompt.
>>>>>>
>>>>> With cygwin 3.1.x I can't find a way to make my program work.
>>>>>
>>>>> When running from within a Cygwin64 terminal the `stty -cooked' 
>>>>> command
>>>>> terminates with exit code 0, but the Java program behaves just as if
>>>>> `stty -cooked' is not in effect: It does not handle single keystrokes
>>>>> immediately.
>>>>>
>>>>> When running from a Windows command prompt I can execute the stty
>>>>> program as \cygwin64\bin\stty. However, when given the '-cooked'
>>>>> argument it complains:
>>>>> /usr/bin/stty: 'standard input': unable to perform all requested 
>>>>> operations
>>>>>
>>>>> - and here as well the Java program behaves as if `stty -cooked' 
>>>>> is not
>>>>> in effect.
>>>> Some time has passed; I am just curious if anyone may have found a
>>>> solution to the "stty -cooked" issue. With cygwin-3.0.* it was 
>>>> possible
>>>> to have a Java program act on single keystrokes, with cygwin-3.1 I 
>>>> don't
>>>> know how to do it. Any ideas welcome!
>>> Solution 1:
>>> Redesign your java program using JNA with kbhit()/getch() instead of
>>> System.in.available()/System.in.read().
>>>
>>> Solution 2:
>>> Add SetConsoleMode() call with ENABLE_LINE_INPUT flag cleared using 
>>> JNA.
>> Couldn't cygwin clear this flag when it sets up ConPTY while the pty 
>> is in raw mode?
>> Thomas
>>
>>>
>>> Solution 3:
>>> Use a wrappwer instead of stty such as:
>>>
>>> #include <stdio.h>
>>> #include <unistd.h>
>>> #include <termios.h>
>>> #include <string.h>
>>> #include <pthread.h>
>>> #include <sys/wait.h>
>>>
>>> void *fwd(void *param)
>>> {
>>>      FILE *f = (FILE *) param;
>>>      char buf[128];
>>>      int len;
>>>      while (1) {
>>>          if ((len = read(0, buf, sizeof(buf))) <= 0) break;
>>>          if (write(fileno(f), buf, len) < len) break;
>>>      }
>>>      return NULL;
>>> }
>>>
>>> int main(int argc, char *argv[])
>>> {
>>>      FILE *f;
>>>      int i;
>>>      pthread_t th;
>>>      struct termios t, t_orig;
>>>      char cmd[1024] = {0, };
>>>      if (argc < 2) return 0;
>>>      for (i = 1; i < argc && strlen(cmd)+strlen(argv[i]) < 
>>> sizeof(cmd)-2; i++) {
>>>          sprintf(cmd + strlen(cmd), (i>1)?" %s":"%s", argv[i]);
>>>      }
>>>      f = popen(cmd, "w");
>>>      tcgetattr(0, &t_orig);
>>>      t = t_orig;
>>>      cfmakeraw(&t);
>>>      tcsetattr(0, TCSANOW, &t);
>>>      pthread_create(&th, NULL, fwd, f);
>>>      wait(NULL);
>>>      tcsetattr(0, TCSANOW, &t_orig);
>>>      pclose(f);
>>>      return 0;
>>> }
>>>
>>
>>
> Thanks Takashi for suggesting many ways to solve this! The C wrapper 
> solution looks quite powerful.
Can you please provide a full test case, as it worked for you in cygwin 
3.0.7, i.e. the Java program and the wrapping commands, for some 
experiments.
Thomas

>
> Meanwhile it has occurred to me that bash itself can provide a
> workaround. Since I start the Java console application from a bash
> wrapper already, I have now changed it so that it does something like
> this:
>
> (while true; do read -s -r -N 1 Key; echo "$Key"; done) | java ...
>
> With these options to 'read' I get action on every keystroke, and the
> Java program gets its input as terminated length-1 lines.
> -- 
> Problem reports:      https://cygwin.com/problems.html
> FAQ:                  https://cygwin.com/faq/
> Documentation:        https://cygwin.com/docs.html
> Unsubscribe info:     https://cygwin.com/ml/#unsubscribe-simple


  reply	other threads:[~2020-07-28 23:30 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-01-13 10:53 Rabbe Fogelholm
2020-01-13 11:02 ` Takashi Yano
2020-01-13 11:36   ` Rabbe Fogelholm
2020-07-28 13:56     ` Rabbe Fogelholm
2020-07-28 16:38       ` Takashi Yano
2020-07-28 16:56         ` Thomas Wolff
2020-07-28 17:40           ` Takashi Yano
2020-07-28 19:51           ` Rabbe Fogelholm
2020-07-28 23:30             ` Thomas Wolff [this message]
2020-07-29 20:03               ` Rabbe Fogelholm
2020-07-31  6:52                 ` Rabbe Fogelholm
2020-07-29  0:02             ` Takashi Yano

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=f935b0f6-1de3-e2d1-285e-f26cd945985a@towo.net \
    --to=towo@towo.net \
    --cc=cygwin@cygwin.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).